: This is highly technical and 'geeky' for those who like this kind of thing..
A while ago I approached Sarah Sands at GOLDEN and asked if they would consider releasing the raw CIELAB data from their lightfast testing for their paints (both before and after the accelerated lightfastness testing). Sarah agreed and after some time was able to collect all this information and send it to me, as well as post it on their website:“You can also access this,and a PDF version, if you go to our Tech Info page
and scroll to just under the Application Sheets section, where you will see the Accelerated Lightfastness Data”
GOLDEN have already published their CIELAB data for their heavy body paints in masstone form, while the data before and after lightfast testing are based on averaged readings of 40% reflectance tints following ASTM D4303 lightfastness testing protocols.
With the raw CIELAB data I was able to do several things:
- Convert them to RGB and show the colours in masstone, before, and after lightfast testing.
- Convert to HSL and be able to sort by hue, saturation or value (actually using value from the L* component of the CIELAB data).
- Measure the lightfastness according to the DE* formula as used by the ATSM ratings, and by the improved DE2000 formula. I was also able to add my own system ‘value weighted’ which combines the DE2000 rating with a weighting as to how much difference there is in perceptual lightness after the accelerated testing. I prioritise this as changes in lightness can affect a value structure of a painting and our eyes are most sensitive to value differences (see black and white photography), then hue changes, and lastly changes in saturation.
I have now created the attached spreadsheet where you can filter on colour groups, single pigments or mixtures, and sort them according to the formula of your choice (or by hue, saturation, lightness, etc..)
The following observations were made using this modified weighting system with single pigments only:
Most lightfast single pigment: I was expecting an iron oxide or a black, but the most lightfast single pigment in all measuring approaches is.. Graphite Grey (PBk10). A shame that its shiny quality makes it not useful as a general purpose pigment.Red
Some surprises here. After the rock solid Iron oxides of PR101, comes the Red Cadmiums (PR108). However while the light and dark versions do very well, the cadmium red medium shows noticeable increases in its lightness after testing. Is this due to a difference in the pigment batch, the production method, or the testing? Not sure.
Another surprise to me is that Pyrrole Red (PR254) performs significantly and noticeably better than Pyrrole Red Light (PR255). Pyrrole Red Dark (PR264) comes bottom of the list with the largest change in lightness of all the single pigment reds here.Orange
These are not many choices in this colour group. Quinacridone / Nickel Azo Gold (PR206) comes top of the list in DE, DE2000, and the value weighted DE2000, which is a bit surprising as it’s a Quinacridone which are not always the most lightfast. However it’s dark valued so may not be useful for bright saturated oranges. Cadmium Orange (PO20) is second in the list, followed by Perinone Orange (PO43) and Pyrrole Orange (PO73) with more noticeable lightness changes as you go down the list.Yellows
Big surprises here.. Between Raw Sienna (PY43), Mars Yellow (PBr6) and Yellow Oxide (PY42) is Nickel Azo Yellow (PY150) which is excellent. This is dark valued in masstone so might be difficult to use as a primary yellow though.
Titanate Yellow (PY53) also does excellently which I am not so surprised about being a metal complex colour, I expect its cousin chrome titanium oxide (PBr24) will also perform as well but it’s not yet offered by Golden in their Acrylic range as a single pigment colour.
So here is another surprise. In the testing that Golden did, their Diarlylide Yellow HR (PY83) which is often maligned, did better than all the other synthetic yellows (and including cadmiums). It’s an orange leaning yellow which could be an issue for some people who prefer a middle or green leaning yellow.
Then comes Benzimidazolone Yellow Light (PY175) which although transparent has the same hue range as the Cadmium Yellow Medium.
Following these comes the Cadmium Yellow Dark and Medium which perform well. And here is another concern, the Cadmium Yellow Primrose and Cadmium Yellow Light do much worse. I think Handprint.com refers to the worse lightfastness of the more green leaning yellows as well. Both of them darken slightly but the Cadmium Yellow Light also seem to increase in saturation which is not what I expected.. Not sure why this is happening.
The Hansa Yellows follow near the bottom with PY74 being the best of the group but with a slight darkening and changes in hue and saturation. PY3 and PY73 are at the bottom (PY73 being the worst) and both lose about 33% of their saturation which suggests to me that they should not be used and may well fade away much more after more UV exposure.Green
There are not many choices in this colour group, but Phthalo Greens (PG7), Chromium Oxide Greens (PG17) and Cobalt Green (PG26) are all rock solid except that the dark variant of PG17 performs quite a bit worse in comparison to the others.Blue
Prussian Blue (PB27) is not available in acrylics, but of the blues tested the Cerulean Blues of PB36:1 and PB36 are rock solid. Ultramarine Blue (PB29) is also excellent as are the phthalo blues (PB15:0 and PB15:3). The cobalts (PB28 and PG36) fair a little worse. Anthraquione Blue (PB60) performs at the bottom of the list with a noticeable increase in its lightness after testing. I believe this is a vat dye, so perhaps not surprising it doesn’t do as well as some others.Violet / Purple
Some of the colours in this group are reddish, but tint towards purple and violet. Ultramarine violet (PV15) is the best in terms of lightfastness, with the Quinacridones (PV19 and PV122) showing some small changes after testing. The Violet learning PV19 seems to be the best of these.
Dioxazine Purple is by far the worst showing significant changes in hue, saturation, and value. So much so that I would not recommend this to be used, instead PV15 is a good choice, although it’s a weak tinter. GOLDEN don’t have a PB29 / PR122 mix, but they do have a permanent violet dark (PB60 / PR122) which performs very well. That suggests that a PB29/PR122 or PB29 / PV19 mix would be a good choice too as PB29 (or PB15:0) are more lightfast blues.A few observations on mixtures:
Alizarin Crimson Hue (PR122 / PR206 / PG7) is excellent, and is rock solid.
Napthol Reds are poor and come towards the bottom of the list.
Any mixture with PV23 suffers.
Titanium Buff (PW6.1 / PW6) does really badly. Not sure why this is, but it suggests mixing PR101 or PBr7 with PW6 would be a better option.
Hope this proves useful
(see the attached lightfastness spreadsheet)